nominatim/tokenizer/sanitizers/tag_japanese.py

   1 # SPDX-License-Identifier: GPL-3.0-or-later
   2 #
   3 # This file is part of Nominatim. (https://nominatim.org)
   4 #
   5 # Copyright (C) 2022 by the Nominatim developer community.
   6 # For a full list of authors see the git log.
   7 """
   8 This sanitizer maps OSM data to Japanese block addresses.
   9 It replaces blocknumber and housenumber with housenumber,
  10 and quarter and neighbourhood with place.
  11 """
  12
  13
  14 from typing import Callable
  15 from typing import List, Optional
  16
  17 from nominatim.tokenizer.sanitizers.base import ProcessInfo
  18 from nominatim.tokenizer.sanitizers.config import SanitizerConfig
  19 from nominatim.data.place_name import PlaceName
  20
  21 KANJI_MAP = {
  22       '零': '0',
  23       '一': '1',
  24       '二': '2',
  25       '三': '3',
  26       '四': '4',
  27       '五': '5',
  28       '六': '6',
  29       '七': '7',
  30       '八': '8',
  31       '九': '9'
  32     }
  33
  34 def convert_kanji_sequence_to_number(sequence: str) -> str:
  35     """Converts Kanji numbers to Arabic numbers
  36     """
  37     converted = ''
  38     current_number = ''
  39     for char in sequence:
  40         if char in KANJI_MAP:
  41             current_number += KANJI_MAP[char]
  42         else:
  43             converted += current_number
  44             current_number = ''
  45             converted += char
  46     converted += current_number
  47     return converted
  48
  49 def create(_: SanitizerConfig) -> Callable[[ProcessInfo], None]:
  50 #def create(config: SanitizerConfig) -> Callable[[ProcessInfo],None]:
  51     """Set up the sanitizer
  52     """
  53     return tag_japanese
  54     #return tag_japanese(config)
  55
  56 def reconbine_housenumber(
  57     new_address: List[PlaceName],
  58     tmp_housenumber: Optional[str],
  59     tmp_blocknumber: Optional[str]
  60 ) -> List[PlaceName]:
  61     """ Recombine the tag of housenumber by using housenumber and blocknumber
  62     """
  63     if tmp_blocknumber and tmp_housenumber:
  64         new_address.append(
  65             PlaceName(
  66                 kind='housenumber',
  67                 name=f'{tmp_blocknumber}-{tmp_housenumber}',
  68                 suffix=''
  69             )
  70         )
  71     elif tmp_blocknumber:
  72         new_address.append(
  73             PlaceName(
  74                 kind='housenumber',
  75                 name=f'{tmp_blocknumber}',
  76                 suffix=''
  77             )
  78         )
  79     elif tmp_housenumber:
  80         new_address.append(
  81             PlaceName(
  82                 kind='housenumber',
  83                 name=f'{tmp_housenumber}',
  84                 suffix=''
  85             )
  86         )
  87     return new_address
  88
  89 def reconbine_place(
  90     new_address: List[PlaceName],
  91     tmp_neighbourhood: Optional[str],
  92     tmp_quarter: Optional[str]
  93 ) -> List[PlaceName]:
  94     """ Recombine the tag of place by using neighbourhood and quarter
  95     """
  96     if tmp_neighbourhood and tmp_quarter:
  97         new_address.append(
  98             PlaceName(
  99                 kind='place',
 100                 name=f'{tmp_quarter}{tmp_neighbourhood}',
 101                 suffix=''
 102             )
 103         )
 104     elif tmp_neighbourhood:
 105         new_address.append(
 106             PlaceName(
 107                 kind='place',
 108                 name=f'{tmp_neighbourhood}',
 109                 suffix=''
 110             )
 111         )
 112     elif tmp_quarter:
 113         new_address.append(
 114             PlaceName(
 115                 kind='place',
 116                 name=f'{tmp_quarter}',
 117                 suffix=''
 118             )
 119         )
 120     return new_address
 121 def tag_japanese(obj: ProcessInfo) -> None:
 122     """Recombine kind of address
 123     """
 124     if obj.place.country_code != 'jp':
 125         return
 126     tmp_housenumber = None
 127     tmp_blocknumber = None
 128     tmp_neighbourhood = None
 129     tmp_quarter = None
 130
 131     new_address = []
 132     for item in obj.names:
 133         item.name = convert_kanji_sequence_to_number(item.name)
 134
 135     for item in obj.address:
 136         item.name = convert_kanji_sequence_to_number(item.name)
 137         if item.kind == 'housenumber':
 138             tmp_housenumber = item.name
 139         elif item.kind == 'block_number':
 140             tmp_blocknumber = item.name
 141         elif item.kind == 'neighbourhood':
 142             tmp_neighbourhood = item.name
 143         elif item.kind == 'quarter':
 144             tmp_quarter = item.name
 145         else:
 146             new_address.append(item)
 147
 148     new_address = reconbine_housenumber(new_address,tmp_housenumber,tmp_blocknumber)
 149     new_address = reconbine_place(new_address,tmp_neighbourhood,tmp_quarter)
 150
 151     obj.address = [item for item in new_address if item.name is not None]